D-Net
/
dnet-hadoop
14
5
Fork 11
Code Issues 14 Pull Requests 13 Projects 2 Releases 2 Wiki Activity

Merge pull request 'EOSC IF' (#230) from tagEosc into beta 

Reviewed-on: #230
This commit is contained in:
Claudio Atzori 2022-07-25 14:14:53 +02:00
parent 3f883c4ecc 612b7a5530
commit 3c23d634eb
3 changed files with 235 additions and 75 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

98
dhp-workflows/dhp-enrichment/src/main/java/eu/dnetlib/dhp/bulktag/SparkEoscTag.java
View File

@ -28,28 +28,6 @@ import eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils;
public class SparkEoscTag {
private static final Logger log = LoggerFactory.getLogger(SparkEoscTag.class);
public static final ObjectMapper OBJECT_MAPPER = new ObjectMapper();
public static final Qualifier EOSC_QUALIFIER = OafMapperUtils
.qualifier(
"EOSC",
"European Open Science Cloud",
ModelConstants.DNET_SUBJECT_TYPOLOGIES, ModelConstants.DNET_SUBJECT_TYPOLOGIES);
public static final DataInfo EOSC_DATAINFO = OafMapperUtils
.dataInfo(
false, "propagation", true, false,
OafMapperUtils
.qualifier(
"propagation:subject", "Inferred by OpenAIRE",
ModelConstants.DNET_PROVENANCE_ACTIONS, ModelConstants.DNET_PROVENANCE_ACTIONS),
"0.9");
public final static StructuredProperty EOSC_NOTEBOOK = OafMapperUtils
.structuredProperty(
"EOSC::Jupyter Notebook", EOSC_QUALIFIER, EOSC_DATAINFO);
public final static StructuredProperty EOSC_GALAXY = OafMapperUtils
.structuredProperty(
"EOSC::Galaxy Workflow", EOSC_QUALIFIER, EOSC_DATAINFO);
public final static StructuredProperty EOSC_TWITTER = OafMapperUtils
.structuredProperty(
"EOSC::Twitter Data", EOSC_QUALIFIER, EOSC_DATAINFO);
public static void main(String[] args) throws Exception {
String jsonConfiguration = IOUtils
@ -84,29 +62,35 @@ public class SparkEoscTag {
});
}
public static EoscIfGuidelines newInstance(String code, String label, String url, String semantics) {
EoscIfGuidelines eig = new EoscIfGuidelines();
eig.setCode(code);
eig.setLabel(label);
eig.setUrl(url);
eig.setSemanticRelation(semantics);
return eig;
}
private static void execEoscTag(SparkSession spark, String inputPath, String workingPath) {
readPath(spark, inputPath + "/software", Software.class)
.map((MapFunction<Software, Software>) s -> {
List<StructuredProperty> sbject;
if (!Optional.ofNullable(s.getSubject()).isPresent())
s.setSubject(new ArrayList<>());
sbject = s.getSubject();
if (containsCriteriaNotebook(s)) {
sbject.add(EOSC_NOTEBOOK);
if (sbject.stream().anyMatch(sb -> sb.getValue().equals("EOSC Jupyter Notebook"))) {
sbject = sbject.stream().map(sb -> {
if (sb.getValue().equals("EOSC Jupyter Notebook")) {
return null;
}
return sb;
}).filter(Objects::nonNull).collect(Collectors.toList());
s.setSubject(sbject);
}
if (!Optional.ofNullable(s.getEoscifguidelines()).isPresent())
s.setEoscifguidelines(new ArrayList<>());
addEIG(
s.getEoscifguidelines(), "EOSC::Jupyter Notebook", "EOSC::Jupyter Notebook", "",
"compliesWith");
}
if (containsCriteriaGalaxy(s)) {
sbject.add(EOSC_GALAXY);
if (!Optional.ofNullable(s.getEoscifguidelines()).isPresent())
s.setEoscifguidelines(new ArrayList<>());
addEIG(
s.getEoscifguidelines(), "EOSC::Galaxy Workflow", "EOSC::Galaxy Workflow", "", "compliesWith");
}
return s;
}, Encoders.bean(Software.class))
@ -123,15 +107,17 @@ public class SparkEoscTag {
readPath(spark, inputPath + "/otherresearchproduct", OtherResearchProduct.class)
.map((MapFunction<OtherResearchProduct, OtherResearchProduct>) orp -> {
List<StructuredProperty> sbject;
if (!Optional.ofNullable(orp.getSubject()).isPresent())
orp.setSubject(new ArrayList<>());
sbject = orp.getSubject();
if (!Optional.ofNullable(orp.getEoscifguidelines()).isPresent())
orp.setEoscifguidelines(new ArrayList<>());
if (containsCriteriaGalaxy(orp)) {
sbject.add(EOSC_GALAXY);
addEIG(
orp.getEoscifguidelines(), "EOSC::Galaxy Workflow", "EOSC::Galaxy Workflow", "",
"compliesWith");
}
if (containscriteriaTwitter(orp)) {
sbject.add(EOSC_TWITTER);
addEIG(orp.getEoscifguidelines(), "EOSC::Twitter Data", "EOSC::Twitter Data", "", "compliesWith");
}
return orp;
}, Encoders.bean(OtherResearchProduct.class))
@ -148,12 +134,11 @@ public class SparkEoscTag {
readPath(spark, inputPath + "/dataset", Dataset.class)
.map((MapFunction<Dataset, Dataset>) d -> {
List<StructuredProperty> sbject;
if (!Optional.ofNullable(d.getSubject()).isPresent())
d.setSubject(new ArrayList<>());
sbject = d.getSubject();
if (!Optional.ofNullable(d.getEoscifguidelines()).isPresent())
d.setEoscifguidelines(new ArrayList<>());
if (containscriteriaTwitter(d)) {
sbject.add(EOSC_TWITTER);
addEIG(d.getEoscifguidelines(), "EOSC::Twitter Data", "EOSC::Twitter Data", "", "compliesWith");
}
return d;
}, Encoders.bean(Dataset.class))
@ -169,6 +154,12 @@ public class SparkEoscTag {
.json(inputPath + "/dataset");
}
private static void addEIG(List<EoscIfGuidelines> eoscifguidelines, String code, String label, String url,
String sem) {
if (!eoscifguidelines.stream().anyMatch(eig -> eig.getCode().equals(code)))
eoscifguidelines.add(newInstance(code, label, url, sem));
}
private static boolean containscriteriaTwitter(Result r) {
Set<String> words = getWordsSP(r.getTitle());
words.addAll(getWordsF(r.getDescription()));
@ -212,13 +203,6 @@ public class SparkEoscTag {
return false;
}
private static Set<String> getSubjects(List<StructuredProperty> s) {
Set<String> subjects = new HashSet<>();
s.stream().forEach(sbj -> subjects.addAll(Arrays.asList(sbj.getValue().toLowerCase().split(" "))));
s.stream().forEach(sbj -> subjects.add(sbj.getValue().toLowerCase()));
return subjects;
}
private static Set<String> getWordsSP(List<StructuredProperty> elem) {
Set<String> words = new HashSet<>();
Optional
@ -242,9 +226,7 @@ public class SparkEoscTag {
t -> words
.addAll(
Arrays.asList(t.getValue().toLowerCase().replaceAll("[^a-zA-Z ]", "").split(" ")))));
// elem
// .forEach(
// t -> words.addAll(Arrays.asList(t.getValue().toLowerCase().replaceAll("[^a-zA-Z ]", "").split(" "))));
return words;
}

210
dhp-workflows/dhp-enrichment/src/test/java/eu/dnetlib/dhp/bulktag/EOSCTagJobTest.java
View File

@ -126,10 +126,23 @@ public class EOSCTagJobTest {
.assertEquals(
4,
tmp
.filter(s -> s.getEoscifguidelines() != null)
.filter(
s -> s.getSubject().stream().anyMatch(sbj -> sbj.getValue().equals("EOSC::Jupyter Notebook")))
s -> s
.getEoscifguidelines()
.stream()
.anyMatch(eig -> eig.getCode().equals("EOSC::Jupyter Notebook")))
.count());
Assertions
.assertEquals(
1, tmp
.filter(sw -> sw.getId().equals("50|od______1582::4132f5ec9496f0d6adc7b00a50a56ff4"))
.collect()
.get(0)
.getEoscifguidelines()
.size());
Assertions
.assertEquals(
1, tmp
@ -140,6 +153,16 @@ public class EOSCTagJobTest {
.size());
Assertions
.assertTrue(
tmp
.filter(sw -> sw.getId().equals("50|od______1582::4132f5ec9496f0d6adc7b00a50a56ff4"))
.collect()
.get(0)
.getEoscifguidelines()
.stream()
.anyMatch(s -> s.getCode().equals("EOSC::Jupyter Notebook")));
Assertions
.assertFalse(
tmp
.filter(sw -> sw.getId().equals("50|od______1582::4132f5ec9496f0d6adc7b00a50a56ff4"))
.collect()
@ -166,16 +189,24 @@ public class EOSCTagJobTest {
.stream()
.anyMatch(s -> s.getValue().equals("EOSC::Jupyter Notebook")));
Assertions
.assertTrue(
tmp
.filter(sw -> sw.getId().equals("50|od______1582::501b25d420f808c8eddcd9b16e917f11"))
.collect()
.get(0)
.getEoscifguidelines() == null);
Assertions
.assertEquals(
9, tmp
8, tmp
.filter(sw -> sw.getId().equals("50|od______1582::581621232a561b7e8b4952b18b8b0e56"))
.collect()
.get(0)
.getSubject()
.size());
Assertions
.assertTrue(
.assertFalse(
tmp
.filter(sw -> sw.getId().equals("50|od______1582::581621232a561b7e8b4952b18b8b0e56"))
.collect()
@ -183,6 +214,23 @@ public class EOSCTagJobTest {
.getSubject()
.stream()
.anyMatch(s -> s.getValue().equals("EOSC::Jupyter Notebook")));
Assertions
.assertEquals(
1, tmp
.filter(sw -> sw.getId().equals("50|od______1582::581621232a561b7e8b4952b18b8b0e56"))
.collect()
.get(0)
.getEoscifguidelines()
.size());
Assertions
.assertTrue(
tmp
.filter(sw -> sw.getId().equals("50|od______1582::581621232a561b7e8b4952b18b8b0e56"))
.collect()
.get(0)
.getEoscifguidelines()
.stream()
.anyMatch(s -> s.getCode().equals("EOSC::Jupyter Notebook")));
Assertions
.assertEquals(
@ -201,17 +249,24 @@ public class EOSCTagJobTest {
.getSubject()
.stream()
.anyMatch(s -> s.getValue().equals("EOSC::Jupyter Notebook")));
Assertions
.assertTrue(
tmp
.filter(sw -> sw.getId().equals("50|od______1582::5aec1186054301b66c0c5dc35972a589"))
.collect()
.get(0)
.getEoscifguidelines() == null);
Assertions
.assertEquals(
9, tmp
8, tmp
.filter(sw -> sw.getId().equals("50|od______1582::639909adfad9d708308f2aedb733e4a0"))
.collect()
.get(0)
.getSubject()
.size());
Assertions
.assertTrue(
.assertFalse(
tmp
.filter(sw -> sw.getId().equals("50|od______1582::639909adfad9d708308f2aedb733e4a0"))
.collect()
@ -219,14 +274,31 @@ public class EOSCTagJobTest {
.getSubject()
.stream()
.anyMatch(s -> s.getValue().equals("EOSC::Jupyter Notebook")));
Assertions
.assertEquals(
1,
tmp
.filter(sw -> sw.getId().equals("50|od______1582::639909adfad9d708308f2aedb733e4a0"))
.collect()
.get(0)
.getEoscifguidelines()
.size());
Assertions
.assertTrue(
tmp
.filter(sw -> sw.getId().equals("50|od______1582::639909adfad9d708308f2aedb733e4a0"))
.collect()
.get(0)
.getEoscifguidelines()
.stream()
.anyMatch(s -> s.getCode().equals("EOSC::Jupyter Notebook")));
List<StructuredProperty> subjects = tmp
.filter(sw -> sw.getId().equals("50|od______1582::6e7a9b21a2feef45673890432af34244"))
.collect()
.get(0)
.getSubject();
Assertions.assertEquals(8, subjects.size());
Assertions.assertTrue(subjects.stream().anyMatch(s -> s.getValue().equals("EOSC::Jupyter Notebook")));
Assertions.assertEquals(7, subjects.size());
Assertions.assertTrue(subjects.stream().anyMatch(s -> s.getValue().equals("jupyter")));
Assertions.assertTrue(subjects.stream().anyMatch(s -> s.getValue().equals("Modeling and Simulation")));
Assertions.assertTrue(subjects.stream().anyMatch(s -> s.getValue().equals("structure granulaire")));
@ -250,6 +322,17 @@ public class EOSCTagJobTest {
.filter(
ds -> ds.getSubject().stream().anyMatch(sbj -> sbj.getValue().equals("EOSC::Jupyter Notebook")))
.count());
Assertions
.assertEquals(
0, sc
.textFile(workingDir.toString() + "/input/dataset")
.map(item -> OBJECT_MAPPER.readValue(item, Dataset.class))
.filter(
ds -> ds
.getEoscifguidelines()
.stream()
.anyMatch(eig -> eig.getCode().equals("EOSC::Jupyter Notebook")))
.count());
Assertions
.assertEquals(
@ -264,7 +347,22 @@ public class EOSCTagJobTest {
.textFile(workingDir.toString() + "/input/otherresearchproduct")
.map(item -> OBJECT_MAPPER.readValue(item, OtherResearchProduct.class))
.filter(
ds -> ds.getSubject().stream().anyMatch(sbj -> sbj.getValue().equals("EOSC::Jupyter Notebook")))
orp -> orp
.getSubject()
.stream()
.anyMatch(sbj -> sbj.getValue().equals("EOSC::Jupyter Notebook")))
.count());
Assertions
.assertEquals(
0, sc
.textFile(workingDir.toString() + "/input/otherresearchproduct")
.map(item -> OBJECT_MAPPER.readValue(item, OtherResearchProduct.class))
.filter(
orp -> orp
.getSubject()
.stream()
.anyMatch(eig -> eig.getValue().equals("EOSC::Jupyter Notebook")))
.count());
// spark.stop();
@ -326,22 +424,41 @@ public class EOSCTagJobTest {
Assertions
.assertEquals(
1,
0,
tmp
.filter(
s -> s.getSubject().stream().anyMatch(sbj -> sbj.getValue().equals("EOSC::Galaxy Workflow")))
.count());
Assertions
.assertEquals(
1,
tmp
.filter(
s -> s.getEoscifguidelines() != null)
.count());
Assertions
.assertEquals(
1,
tmp
.filter(
s -> s.getEoscifguidelines() != null)
.filter(
s -> s
.getEoscifguidelines()
.stream()
.anyMatch(eig -> eig.getCode().equals("EOSC::Galaxy Workflow")))
.count());
Assertions
.assertEquals(
2, tmp
1, tmp
.filter(sw -> sw.getId().equals("50|od______1582::4132f5ec9496f0d6adc7b00a50a56ff4"))
.collect()
.get(0)
.getSubject()
.size());
Assertions
.assertTrue(
.assertFalse(
tmp
.filter(sw -> sw.getId().equals("50|od______1582::4132f5ec9496f0d6adc7b00a50a56ff4"))
.collect()
@ -350,6 +467,24 @@ public class EOSCTagJobTest {
.stream()
.anyMatch(s -> s.getValue().equals("EOSC::Galaxy Workflow")));
Assertions
.assertEquals(
1, tmp
.filter(sw -> sw.getId().equals("50|od______1582::4132f5ec9496f0d6adc7b00a50a56ff4"))
.collect()
.get(0)
.getEoscifguidelines()
.size());
Assertions
.assertTrue(
tmp
.filter(sw -> sw.getId().equals("50|od______1582::4132f5ec9496f0d6adc7b00a50a56ff4"))
.collect()
.get(0)
.getEoscifguidelines()
.stream()
.anyMatch(eig -> eig.getCode().equals("EOSC::Galaxy Workflow")));
Assertions
.assertEquals(
5, tmp
@ -385,22 +520,34 @@ public class EOSCTagJobTest {
Assertions
.assertEquals(
1,
0,
orp
.filter(
s -> s.getSubject().stream().anyMatch(sbj -> sbj.getValue().equals("EOSC::Galaxy Workflow")))
.count());
orp.foreach(o -> System.out.println(OBJECT_MAPPER.writeValueAsString(o)));
Assertions
.assertEquals(
3, orp
1, orp
.filter(o -> o.getEoscifguidelines() != null)
.filter(
o -> o
.getEoscifguidelines()
.stream()
.anyMatch(eig -> eig.getCode().equals("EOSC::Galaxy Workflow")))
.count());
Assertions
.assertEquals(
2, orp
.filter(sw -> sw.getId().equals("50|od______2017::0750a4d0782265873d669520f5e33c07"))
.collect()
.get(0)
.getSubject()
.size());
Assertions
.assertTrue(
.assertFalse(
orp
.filter(sw -> sw.getId().equals("50|od______2017::0750a4d0782265873d669520f5e33c07"))
.collect()
@ -408,6 +555,23 @@ public class EOSCTagJobTest {
.getSubject()
.stream()
.anyMatch(s -> s.getValue().equals("EOSC::Galaxy Workflow")));
Assertions
.assertEquals(
1, orp
.filter(sw -> sw.getId().equals("50|od______2017::0750a4d0782265873d669520f5e33c07"))
.collect()
.get(0)
.getEoscifguidelines()
.size());
Assertions
.assertTrue(
orp
.filter(sw -> sw.getId().equals("50|od______2017::0750a4d0782265873d669520f5e33c07"))
.collect()
.get(0)
.getEoscifguidelines()
.stream()
.anyMatch(s -> s.getCode().equals("EOSC::Galaxy Workflow")));
Assertions
.assertEquals(
@ -516,10 +680,20 @@ public class EOSCTagJobTest {
Assertions
.assertEquals(
3,
0,
orp
.filter(s -> s.getSubject().stream().anyMatch(sbj -> sbj.getValue().equals("EOSC::Twitter Data")))
.count());
Assertions
.assertEquals(
3,
orp
.filter(
s -> s
.getEoscifguidelines()
.stream()
.anyMatch(eig -> eig.getCode().equals("EOSC::Twitter Data")))
.count());
JavaRDD<Dataset> dats = sc
.textFile(workingDir.toString() + "/input/dataset")
@ -531,7 +705,11 @@ public class EOSCTagJobTest {
.assertEquals(
3,
dats
.filter(s -> s.getSubject().stream().anyMatch(sbj -> sbj.getValue().equals("EOSC::Twitter Data")))
.filter(
s -> s
.getEoscifguidelines()
.stream()
.anyMatch(eig -> eig.getCode().equals("EOSC::Twitter Data")))
.count());
}

2
pom.xml
View File

@ -801,7 +801,7 @@
<mockito-core.version>3.3.3</mockito-core.version>
<mongodb.driver.version>3.4.2</mongodb.driver.version>
<vtd.version>[2.12,3.0)</vtd.version>
<dhp-schemas.version>[2.12.0]</dhp-schemas.version>
<dhp-schemas.version>[2.12.2-SNAPSHOT]</dhp-schemas.version>
<dnet-actionmanager-api.version>[4.0.3]</dnet-actionmanager-api.version>
<dnet-actionmanager-common.version>[6.0.5]</dnet-actionmanager-common.version>
<dnet-openaire-broker-common.version>[3.1.6]</dnet-openaire-broker-common.version>